| Title: | Statistical Support for Hierarchical Clusters |
|---|---|
| Description: | Generates n hierarchical clustering hypotheses on subsets of classifiers (usually species in community ecology studies). The n clustering hypotheses are combined to generate a generalized cluster, and computes three metrics of support. 1) The average proportion of elements conforming the group in each of the n clusters (integrity). And 2) the contamination, i.e., the average proportion of elements from other groups that enter a focal group. 3) The probability of existence of the group gives the integrity and contamination in a Bayesian approach. |
| Authors: | Roger Guevara [aut, cre], Ian MacGregor-Fors [aut] |
| Maintainer: | Roger Guevara <[email protected]> |
| License: | GPL-3 |
| Version: | 0.1.1 |
| Built: | 2024-12-01 08:47:03 UTC |
| Source: | CRAN |
Calculates the complementaritry and dissimilarity between two or more communities base on .
compsim2(M)compsim2(M)
M |
A species times site matrix. |
Generates a disimilarity matrix.
Roger Guevara ([email protected])
Aguirre, A., Guevara, R., García, M., & López, J. C. (2010). Fate of epiphytes on phorophytes with different architectural characteristics along the perturbation gradient of Sabal mexicana forests in Veracruz, Mexico. Journal of Vegetation Science, 21(1), 6-15. https://doi.org/10.1111/j.1654-1103.2009.01131.x
data(dune) M2 <- compsim2(t(dune[1:2,])) M2data(dune) M2 <- compsim2(t(dune[1:2,])) M2
Draws the resulting generalized cluster with tips facing either down, left or right.
draw.octopucs(M2, cex=1, lab.pos=0.09, n=NULL, distance=NULL, method=NULL, prop=NULL, Name=NULL, ylim=NULL, xlim=NULL, verbose=NULL, orientation=NULL, support=NULL, sep.lab=NULL, disp = NULL)draw.octopucs(M2, cex=1, lab.pos=0.09, n=NULL, distance=NULL, method=NULL, prop=NULL, Name=NULL, ylim=NULL, xlim=NULL, verbose=NULL, orientation=NULL, support=NULL, sep.lab=NULL, disp = NULL)
M2 |
A lis object generated by glzcluster or glzcluster2 |
cex |
Size of text in the figure |
lab.pos |
This argument modifies the position of the labels relative to the tips of the dendrogram (sites/conditions). |
n |
This is defined in the glzcluster or glzcluster2 functions. Number of bootstrap samples performed. |
distance |
Distance method used (default="bray"). This is defined in the glzcluster or glzcluster2 functions. Options include those of the function "vegdist" ("vegan"): "manhattan", "euclidean", "canberra", "bray", "kulczynski", "gower", "morisita", "horn", "mountford", "jaccard", "raup", "binomial", "chao", "altGower", "cao" and "mahalanobis", as well as those referred in Koleff et al. (2003), available either by name or number: 1 or "w", 2 or "-1", 3 or "c", 4 or "wb", 5 or "r", 6 or "I", 7 or "e", 8 or "t", 9 or "me", 10 or "j", 11 or "sor", 12 or "m", 13 or "-2", 14 or "co", 15 or "cc", 16 or "g", 17 or "-3", 18 or "1", 19 or "", 20 or "hk", 21 or "rlb", 22 or "sim", 23 or "gl", 24 or "z". For open distance metrics (not restricted to zero-one ranges), the axis is re-scaled to cover a zero-one range. If so, the axis will be labeled a "Similarity (rescaled)"; if no re-scaling is necessary, the label will appear as "Similarity". |
method |
This is defined in the glzcluster or glzcluster2 functions. Agglomerative procedure for hierarchical clustering, including: "average" (=UPGMA), "ward.D", "ward.D2", "single", "complete", "mcquitty" (=WPGMA), "median" (=WPGMC) or "centroid" (=UPGMC). |
prop |
This is defined in the glzcluster or glzcluster2 functions. Proportion of information used in the bootstrapping procedures. When matrices contain more than 18 columns, it refers to: (i) the proportion of columns used in every run to produce a new grouping array when "resample=TRUE" and (ii) the amount of cells in the body of the matrix (1–prop; e.g., 1–0.8=0.2) that will be replaced with column averages when "resample=FALSE". When matrices contain less than 18 columns, the procedure with "resample=FALSE" will be forced. In such scenario, "prop" corresponds to the amount of cells in the body of the matrix (1–prop; e.g., 1–0.8=0.2) that will be replaced in the bootstrapping procedures. |
Name |
Name of the database run on octopucs |
xlim |
Limits of the x-axis |
ylim |
Limits of the y-axis |
verbose |
TRUE = display all relevant setup values used in the analysis. FALSE = display only the dendrogram with marked significantly grouped clusters. |
orientation |
Any of d, l, or r denoting the orientation of the tips (down, left, and right, respectively) |
support |
A matrix object generated by the function glzcs |
sep.lab |
This argument modifies the position of supporting values from the nodes and lines and the dendrogram. |
disp |
What to display in relation to supporting values for the groups and subgroups. By default is set to "sig" and then support values are shown only in those groups statistically supported. |
This function draws and if requested, populate with support metrics the consesnsus hypothesis of clustering
Roger Guevara ([email protected]) & Ian MacGregor-Fors ([email protected])
Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence–absence data. Journal of Animal Ecology 71:367. https://doi.org/10.1046/j.1365-2656.2003.00710.x
data(dune) rownames(dune)<- paste("S",1:20, sep="") OCT <- glzcluster(dune, n=100) GRP <- IDscluster(OCT) SUP <- glzcs(GRP, OCT) draw.octopucs(OCT, support=SUP)data(dune) rownames(dune)<- paste("S",1:20, sep="") OCT <- glzcluster(dune, n=100) GRP <- IDscluster(OCT) SUP <- glzcs(GRP, OCT) draw.octopucs(OCT, support=SUP)
Internal function to label the membership of two clusters
F1(A, B)F1(A, B)
A |
Memberships in the first cluster |
B |
Membership in the second cluster |
Produced a two-column matrix populated with zeros and ones to indicate what IDs (row names) are included in each of two vectors in which some element may be shared
Roger Guevara ([email protected])
Generates a cluster hypothesis based on n subsets of the available columns.
glzcluster(data, n = NULL, prop=NULL, distance =NULL, method=NULL, Name=NULL)glzcluster(data, n = NULL, prop=NULL, distance =NULL, method=NULL, Name=NULL)
data |
A matrix object of r rows that will be grouped by c columns. |
n |
The number of subsets to be generated is based on which the generalized cluster will be generated. |
prop |
Either the proportion of the c columns used in every run to produce a new grouping array or the 1-prop number of cells in the body of the matrix whose values are replaced by the corresponding column average. This will be automatically decided based on the total number of columns in the matrix. When there are fewer than 19 columns in the matrix, the procedure will replace cell values with column averages even if the resample is TRUE. |
distance |
Distance method to be used (default="bray"). Options include those of the function "vegdist" ("vegan"): "manhattan", "euclidean", "canberra", "bray", "kulczynski", "gower", "morisita", "horn", "mountford", "jaccard", "raup", "binomial", "chao", "altGower", "cao" and "mahalanobis", as well as those referred in Koleff et al. (2003), available either by name or number: 1 or "w", 2 or "-1", 3 or "c", 4 or "wb", 5 or "r", 6 or "I", 7 or "e", 8 or "t", 9 or "me", 10 or "j", 11 or "sor", 12 or "m", 13 or "-2", 14 or "co", 15 or "cc", 16 or "g", 17 or "-3", 18 or "1", 19 or "", 20 or "hk", 21 or "rlb", 22 or "sim", 23 or "gl", 24 or "z". The axis is re-scaled to cover a zero-one range for open distance metrics (not restricted to zero-one ranges). If so, the axis will be labeled a "Similarity (rescaled)"; if no re-scaling is necessary, the label will appear as "Similarity". |
method |
Agglomerative procedure for hierarchical clustering, including: "average" (=UPGMA), "ward.D", "ward.D2", "single", "complete", "mcquitty" (=WPGMA), "median" (=WPGMC) or "centroid" (=UPGMC). |
Name |
Name of the object matrix analyzed |
Generates a list object that contains the n cluster assableged with random subsets of the data and in the n+1 possition the generalized cluster, an hclust output
Roger Guevara ([email protected]) & Ian MacGregor-Fors ([email protected])
Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence-absence data. Journal of Animal Ecology 71:367. https://doi.org/10.1046/j.1365-2656.2003.00710.x
data(dune) M2 <- glzcluster(dune, n=100, prop=0.8, distance="bray", method="average") length(M2) ls(M2[[length(M2)]])data(dune) M2 <- glzcluster(dune, n=100, prop=0.8, distance="bray", method="average") length(M2) ls(M2[[length(M2)]])
Generates a cluster hypothesis based on the transformation of the 1-prop ratio of the cells in the matrix's body into their corresponding column averages.
glzcluster2(data=data, n = NULL, prop=NULL, distance =NULL, method=NULL, Name=NULL)glzcluster2(data=data, n = NULL, prop=NULL, distance =NULL, method=NULL, Name=NULL)
data |
A matrix object of r rows that will be gropued by c columns. |
n |
The number of subsets to be generated is based on which the generalized cluster will be generated. |
prop |
Either the proportion of the c columns used in every run to produce a new grouping array or the 1-prop number of cells in the body of the matrix whose values are replaced by the corresponding column average. This will be automatically decided based on the total number of columns in the matrix. When there are less than 19 columns in the matrix, the procedure will replace cell values with column averages even if the resample is TRUE. |
distance |
Distance method to be used (default="bray"). Options include those of the function "vegdist" ("vegan"): "manhattan", "euclidean", "canberra", "bray", "kulczynski", "gower", "morisita", "horn", "mountford", "jaccard", "raup", "binomial", "chao", "altGower", "cao" and "mahalanobis", as well as those referred in Koleff et al. (2003), available either by name or number: 1 or "w", 2 or "-1", 3 or "c", 4 or "wb", 5 or "r", 6 or "I", 7 or "e", 8 or "t", 9 or "me", 10 or "j", 11 or "sor", 12 or "m", 13 or "-2", 14 or "co", 15 or "cc", 16 or "g", 17 or "-3", 18 or "1", 19 or "", 20 or "hk", 21 or "rlb", 22 or "sim", 23 or "gl", 24 or "z". For open distance metrics (not restricted to zero-one ranges), the axis is re-scaled to cover a zero-one range. If so, the axis will be labeled a "Similarity (rescaled)"; if no re-scaling is necessary, the label will appear as "Similarity". |
method |
Agglomerative procedure for hierarchical clustering, including: "average" (=UPGMA), "ward.D", "ward.D2", "single", "complete", "mcquitty" (=WPGMA), "median" (=WPGMC) or "centroid" (=UPGMC). |
Name |
Name of the matrix object to be analyzed |
Generates a list object that contains the n cluster assembled with random subsets of the data, and in the n+1 position the generalized cluster, an hclust output
Roger Guevara ([email protected]) & Ian MacGregor-Fors ([email protected])
Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence-absence data. Journal of Animal Ecology 71:367. https://doi.org/10.1046/j.1365-2656.2003.00710.x
data(dune) M2 <- glzcluster(dune, n=100, prop=0.8, distance="bray", method="average") length(M2) ls(M2[[length(M2)]])data(dune) M2 <- glzcluster(dune, n=100, prop=0.8, distance="bray", method="average") length(M2) ls(M2[[length(M2)]])
Generates three metrics of support for the observed clusters at all hierarchical levels
glzcs(GRPs=GRPs, M2=M2)glzcs(GRPs=GRPs, M2=M2)
GRPs |
List object generated by the functions IDscluster & IDscluster2 |
M2 |
Class list object generated by the functions glzcluster or glzcluster2 |
A table with supporting values and size of every group: integrity, contamination, probability and group size.
Roger Guevara ([email protected]) & Ian MacGregor-Fors ([email protected])
Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence-absence data. Journal of Animal Ecology 71:367. https://doi.org/10.1046/j.1365-2656.2003.00710.x
data(dune) OCT <- glzcluster(dune, n=100) GRP <- IDscluster(OCT) SUP <- glzcs(GRP, OCT) SUPdata(dune) OCT <- glzcluster(dune, n=100) GRP <- IDscluster(OCT) SUP <- glzcs(GRP, OCT) SUP
Identifies the memberships at every level of the topology of the generalized cluster.
IDscluster(M2=M2)IDscluster(M2=M2)
M2 |
List object generated by glzcluster or glzcluster2 |
A list object with the memberships for the groups and subgroups all through the topology of the generalized cluster
Roger Guevara ([email protected]) & Ian MacGregor-Fors ([email protected])
Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence–absence data. Journal of Animal Ecology 71:367
data(dune) M2 <- glzcluster(dune, n=5, prop=0.8, distance="bray", method="average") GRPs <- IDscluster(M2) GRPsdata(dune) M2 <- glzcluster(dune, n=5, prop=0.8, distance="bray", method="average") GRPs <- IDscluster(M2) GRPs
Identify the memberships at every level of the topology in the bootstrapped clusters.
IDscluster2(M2=M2)IDscluster2(M2=M2)
M2 |
List object generated by glzcluster or glzcluster2 |
A list object with the memberships for the groups and subgroups all through the topology of the generalized cluster
Roger Guevara ([email protected]) & Ian MacGregor-Fors ([email protected])
Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence-absence data. Journal of Animal Ecology 71:367. https://doi.org/10.1046/j.1365-2656.2003.00710.x
data(dune) M2 <- glzcluster(dune, n=5, prop=0.8, distance="bray", method="average") GRPs <- IDscluster2(M2[[length(M2)]]) GRPsdata(dune) M2 <- glzcluster(dune, n=5, prop=0.8, distance="bray", method="average") GRPs <- IDscluster2(M2[[length(M2)]]) GRPs
The global function of the "octopucs" package is to estimate support for hierarchical clusters throughout the dendrogram's topology, considering their repeatability, integrity, and contamination based on bootstrapping procedures.
octopucs(data, n=NULL, prop=NULL, resample = TRUE, distance =NULL, method=NULL, orientation=NULL, disp = NULL, lab.pos=NULL, sep.lab=NULL, xlim=NULL, ylim=NULL, verbose=NULL)octopucs(data, n=NULL, prop=NULL, resample = TRUE, distance =NULL, method=NULL, orientation=NULL, disp = NULL, lab.pos=NULL, sep.lab=NULL, xlim=NULL, ylim=NULL, verbose=NULL)
data |
A matrix whose rows (sites/conditions) will be grouped based on the columns (species/elements). |
n |
The number of bootstrap repetitions. |
prop |
Proportion of information used in the bootstrapping procedures. When matrices contain more than 18 columns, it refers to (i) the proportion of columns used in every run to produce a new grouping array when "resample=TRUE" and (ii) the number of cells in the body of the matrix (1–prop; e.g., 1–0.8=0.2) that will be replaced with column averages when "resample=FALSE". When matrices contain less than 18 columns, the procedure with "resample=FALSE" will be forced. In such a scenario, "prop" corresponds to the number of cells in the body of the matrix (1–prop; e.g., 1–0.8=0.2) that will be replaced in the bootstrapping procedures. |
resample |
When TRUE, "octopucs" will select random subsets of columns in every bootstrap run to produce a new grouping array. When FALSE, "octopucs" will replace cells in the body of the matrix with column averages instead of entire columns (see description for "prop" for further details). |
distance |
Distance method to be used (default="bray"). Options include those of the function "vegdist" ("vegan"): "manhattan", "euclidean", "canberra", "bray", "kulczynski", "gower", "morisita", "horn", "mountford", "jaccard", "raup", "binomial", "chao", "altGower", "cao" and "mahalanobis", as well as those referred in Koleff et al. (2003), available either by name or number: 1 or "w", 2 or "-1", 3 or "c", 4 or "wb", 5 or "r", 6 or "I", 7 or "e", 8 or "t", 9 or "me", 10 or "j", 11 or "sor", 12 or "m", 13 or "-2", 14 or "co", 15 or "cc", 16 or "g", 17 or "-3", 18 or "1", 19 or "", 20 or "hk", 21 or "rlb", 22 or "sim", 23 or "gl", 24 or "z". For open distance metrics (not restricted to zero-one ranges), the axis is re-scaled to cover a zero-one range. If so, the axis will be labeled a "Similarity (rescaled)"; if no re-scaling is necessary, the label will appear as "Similarity". |
method |
Agglomerative procedure for hierarchical clustering, including: "average" (=UPGMA), "ward.D", "ward.D2", "single", "complete", "mcquitty" (=WPGMA), "median" (=WPGMC) or "centroid" (=UPGMC). |
orientation |
Orientation of the plot (default="r": right; metric in the x-axis is similarity). "l" = left-wise orientation (metric in the x-axis is dissimilarity). "d" = down-wise orientation (metric in the y-axis is dissimilarity). |
disp |
Result display in the dendrogram (default="sig"; only supported nodes will populated with given results). "all" = all nodes will be populated with support metrics, with significantly supported nodes in blue/red and non-significantly supported nodes in black. |
lab.pos |
This argument modifies the position of the labels relative to the tips of the dendrogram (sites/conditions). |
sep.lab |
This argument modifies the position of supporting values from the nodes and lines in the dendrogram. |
xlim |
Limits of the x-axis. |
ylim |
Limits of the y-axis. |
verbose |
TRUE = display all relevant setup values used in the analysis. FALSE = display only the dendrogram with marked significantly grouped clusters. |
"octopucs" generates a list object that contains:
$gcluster |
hclust output of the generalized clustered dendrogram. |
$support |
Table with the support value for every group and subgroup across the topology of the generalized clustered dendrogram. |
$grps |
List with the memberships of every group and subgroup in the topology. |
Roger Guevara ([email protected]) & Ian MacGregor-Fors ([email protected])
Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence-absence data. Journal of Animal Ecology 71:367. https://doi.org/10.1046/j.1365-2656.2003.00710.x
data(dune) octopucs(dune, n=100, prop=0.8, resample=TRUE,distance="bray", method="average", disp = "sig", verbose=TRUE)data(dune) octopucs(dune, n=100, prop=0.8, resample=TRUE,distance="bray", method="average", disp = "sig", verbose=TRUE)